Font Size: a A A

Color Leukocyte Image Segmentation Based On Composite Gradient Watershed Algorithm

Posted on:2013-07-03Degree:MasterType:Thesis
Country:ChinaCandidate:X JiangFull Text:PDF
GTID:2248330371498965Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
With the rapid development of the computer science and technology, computer is more and more widely used in the field of medicine. Cell image processing plays an important role in medical intelligent diagnosis. Cell segmentation is basis of feature extraction and recognition of cells, which also plays a decisive role on identification of cell. The analysis of the morphological specificity of cell’s nucleus and cytoplasm is done for accuracy cell feature extraction. The segmentation algorithm which segments an accuracy contour of nuclei and cytoplasm is a challenging subject at present. The shapes of cells are so many varieties that the traditional segmentation algorithm is difficult to segment contour of the cell from image. There are so many algorithms of cell segmentation but no such an algorithm can be used to segment all cell. Therefore, according to Different types and uses medical cells, people choose different improvement segmentation algorithm to extraction the feature of segmented aim cell.According to the features of dyeing treated color leukocyte image, the paper is put forward a scheme to segment color cells based on the gradient composite watershed algorithm. The watershed algorithm is a classic image segmentation algorithm. However, traditional watershed algorithm only is used in grayscale or gray-gradient, and which is particularly sensitive to the noise of the image, so it is easy to cause over-segmentation of the cell. Therefore, the watershed algorithm is improved in two ways, one is preprocessing of image to restrain the noise, the other is merging region after watershed segmentation to reduce over-segmentation of cell’s images. The gradient has a decisive effect on the watershed algorithm which is improved with eight-neighbor gradient in this paper. The eight-neighbor gradient algorithm is not only in view of the images’grayscale features but also in view of the spatial features of the images. The algorithm can reduce the noise of cell’s image and eliminate part of over-segmentation.Because of the noise or other factors, there are some small areas in the result image after watershed algorithm segmenting. The initial merger is put forward firstly in this paper. This step can not only reduce the computation time but also increase the segmentation accuracy. Then using the LUV color space chromaticity difference of the cells is merged with similar functions. The experiment has proved the accuracy segment the nucleus and the cytoplasm contour.
Keywords/Search Tags:Color cells, Composite gradient, Watershed algorithm, Cell division
PDF Full Text Request
Related items